Epi methods for case-control study in outbreak investigations

Epi methods
1

I am working on statistical analysis of a large bacterial outbreak that happened at a mud adventure race. This is a cumulative case-control study where we established a case definition and then used survey responses to place people in case and control categories. About 62% of race participants were a case (disease was common in this study).

I plan to use multivariate logistic regression. I am just wondering if anyone has experience with conducting more advanced statistical analysis for outbreak investigation that takes into account different confounders and covariates.

My PI mentioned using log binomial regression but I do not think this is the appropriate analysis as this is not a cohort study. Am I wrong?

In short, I think I have my analysis plan but would love to discuss in more detail with people who are more experienced with this type of analysis.

Thank you in advance for your time and insights!

1 Like
2

Hello,

I guess the question is, what are you trying to estimate with your model? Using a logit link function will yield a prevalence-odds ratio where as a log link function will yield a prevalence ratio.

All the best,

Tim

3

Hi julia - if you sent the questionnaire to everyone that attended the race then this is a retrospective cohort study and you could use negative binomial to produce a risk ratio. Even if not everyone responds.